Re: [uracoli-devel] did I tell you?

[Vasilakes, Keith]

My pleasure Axel,
Feel free to use anything I said, I have told you anything proprietary. 
I'd rather you didn't use our company name ( if you even know it ) :)

As far as improvements I think I have told you everything already, mostly being 
able to adjust the clock to get 115200 baud

The only other thing that comes to mind is sleeping. Sleeping is always 
problematic and knowing if sleep actually worked is difficult. 
I don't know how to improve it though, maybe just dead simple examples?

The best part of Uricoli is its simplicity and obviousness. 

Here's another quote for your blog;
The original Bitcloud development took 2 developers 2 years and resulted in a 
network capable of 30 sensors sending data every 30 seconds.
It would never be able to exceed that level with any amount of development

With uricoli 1 developer implemented a system capable of 130+ devices in 1 
month.

Yes the uricoli solution does not do association or multiple hops or routing, 
but in large systems that will limit your network size severely.
Making large networks requires an absolute minimum of traffic and constrained 
retries. Too much of either will fill all available bandwidth.


-----Original Message-----
From: address@hidden [mailto:address@hidden On Behalf Of Axel Wachtler
Sent: Wednesday, September 26, 2012 4:19 PM
To: address@hidden
Subject: Re: [uracoli-devel] did I tell you?

Hallo Keith,

thank you for your very kind and motivating email. It is good to know that the 
software works in such a quite huge network. Of course we like also to know 
about things to improve.

Would it be Ok for you, if we cite your mail in the blog?
I mean a 130 node network would be a good reference.

Best Regards, Axel




On 26.09.2012 20:32, Vasilakes, Keith wrote:
> Axel, Joerg,
>
> Thanks to you and Uricoli, I was able to finish my project AND have it 
> work J
>
> ( it was done a couple of months ago actually)
>
> We were able to create a network of 130 devices ( every device in 
> house) all talking to a coordinator ( a Digi X4 gateway ) with very 
> little data loss. ( few retries here and there )
>
> We normally run at 30 seconds between messages, but were able to crank 
> it up to 10 seconds before the gateway processor started maxing out 
> it's bandwith!
>
> The network itself was stable.
>
> The code base was so much simpler and smaller than the bit bean Zigbee 
> solution, which fell apart at 40 devices.
>
> This was 40 devices sending a data packet every 30 seconds.
>
> We could get the network up, but as soon as one device lost contact it 
> would start the association process, suck up all the bandwith, and now 
> all the other devices fail so the system crashes like a house of cards.
>
> Uricoli never associates and therefore never takes out the network 
> with retries. It just goes to sleep and tries again in 30 seconds.
>
> We do actually associate the devices with the gateway, the gateway has 
> a list of sensors, and it listens for them for 3x the reporting rate.
>
> As sensors check in, the gateway marks them as connected
>
> At the end of the 'Association' period, the gateway reports to a PC 
> app which and how many sensors reported in. so the PC app can inform the user.
>
> So the sensors just do their thing and the gateway worries about the 
> association list.
>
> We did look into writing an association procedure, but decided it 
> wasn't the best way to do it.
>
> All in all awesome, Uricoli was a perfect fit.
>
> *Keith Vasilakes*
>